JPA Hibertane 单向一对一与共享主键先保存父级 NO REVERSE

Question

I'v been searching internet for answer, but nothing was working for me. There is a lot of topics with similar cases but specific details are different in a way that make them unusable for me.

So I have two tables: t_item and t_item_info:

item_id field from t_item_info table references id field from t_item table. I'm using mysql db and id column form t_item is auto incremented

I need to make unidirectional one-to-one mapping in a specific way. Here are my classes:

@Entity
@Table(name = "t_item")
public class Item {
    @Id
    @GeneratedValue(strategy = GenerationType.IDENTITY)
    private Long id;

    private String name;

    @OneToOne(cascade = CascadeType.ALL, fetch = FetchType.LAZY, optional = false)
    @PrimaryKeyJoinColumn(name = "id", referencedColumnName = "item_id")
    private ItemInfo info;
}

And other one

@Entity
@Table(name = "t_item_info")
public class ItemInfo {
    @Id
    private Long itemId;

    private String descr;
}

So the point is that i need Item object to have a reference to ItemInfo object. NOT The other way!

Item -> ItemInfo   --YES
Item <- ItemInfo   --NO

The other thing is that i need parent ( Item ) id to become id for a child ( ItemInfo )

For example I create Item object with null id and set it's info field with ItemInfo object which also have null id field. Like this:

{
  "id": null,
  "name": "Some name",
  "info": {
    "itemId": null,
    "descr": "some descr"
  }
}

Then when Item object persists hibernate should generate id for parent( Item ) object and set it as itemId field for child( ItemInfo ).

I have been trying to achieve this with different hibernate annotations and I noticed that no matter how hard I tried Hibernate always seems to try to persist child object first . I noticed it in the logs when I turned sql logging on. insert into t_item_info always goes first (and dies because of null id:D)

So the question is: Is it even possible to achieve this and if so what should I change in my code to do so

I hope that what I'm trying to ask makes sens to you given my poor explanations =)

Answer 1

Why people always insist the child object table in one-to-one associations should be the one with the foreign key is beyond me.

Anyway, as a workaround, since both objects share the id and the association is non-optional, you might as well declare the autogenerated key for the child object:

@Entity
@Table(name = "t_item_info")
public class ItemInfo {
    @Id
    @GeneratedValue(strategy = GenerationType.IDENTITY)
    private Long itemId;

    private String descr;
}

and then use @MapsId for the parent:

@Entity
@Table(name = "t_item")
public class Item {
    @Id
    private Long id;

    private String name;

    @OneToOne(cascade = CascadeType.ALL, fetch = FetchType.LAZY, optional = false)
    @JoinColumn(name = "id")
    @MapsId
    private ItemInfo info;
}

Note that this approach, will, in a sense, fool Hibernate into thinking it is the Item that should be treated as the child object. You have been warned.

Answer 2

While there is an accepted answer here it looks to me like @Secondary table would be a better and more convenient solution: you may have 2 tables at the database level but I do not see any reason that that fact needs be exposed to any client code. There does not seem to be a lot of benefit to that? The following gives you a simpler API.

Entity:

@Entity
@Table(name = "t_item")
@SecondaryTable("t_item_info",  pkJoinColumns={
        @PrimaryKeyJoinColumn(name="id", referencedColumnName="item_id")})
public class Item {
    @Id
    @GeneratedValue(strategy = GenerationType.IDENTITY)
    private Long id;

    private String name;

    @Colulumn(name = "description", table= "t_item_info"")
    private String description;
}

API:

{
  "id": null,
  "name": "Some name",
  "descr": "some descr"
}